Transforms complex documents like PDFs and Office docs into LLM-ready markdown/JSON for your Agentic workflows.
Python ai4sciencedocument-analysisdocxextract-datalayout-analysisocrparserpdfpdf-converterpdf-extractor-llmpdf-extractor-pretrainpdf-extractor-ragpdf-parserpptxpythonxlsx
Add parameterization for auto mode in PDF analysis tests and enhance remote parse inference progress test
myhloli54373d6fa863
diff --git a/tests/unittest/test_pdf_analyze_block_content.py b/tests/unittest/test_pdf_analyze_block_content.pyindex d9fa9553..b1ac158b 100644--- a/tests/unittest/test_pdf_analyze_block_content.py+++ b/tests/unittest/test_pdf_analyze_block_content.py@@ -129,6 +129,7 @@ def test_convert_vlm_results_deep_copies_projected_mutable_fields() -> None:assert source_cell_merge == [1, 0][email protected]("auto_mode", [False, True])@pytest.mark.parametrize(("effort", "parse_mode", "vlm_method_name"),[@@ -143,6 +144,7 @@ def test_doc_analyze_converts_vlm_results_before_downstream_processing(effort: str,parse_mode: str,vlm_method_name: str,+ auto_mode: bool,) -> None:"""验证 high/xhigh 的 TXT/OCR 路径都只向后续阶段传递原生 list/dict。"""source_block = VlmContentBlock(@@ -154,6 +156,7 @@ def test_doc_analyze_converts_vlm_results_before_downstream_processing(source_page = ExtractResult([source_block], layout_scored=object())fake_document = MagicMock()fake_document.page_count = 1+ fake_document.classify.return_value = parse_modefake_document.__getitem__.return_value = MagicMock(size=(612.0, 792.0))hybrid_model = MagicMock()@@ -222,9 +225,11 @@ def test_doc_analyze_converts_vlm_results_before_downstream_processing(middle_json, model_json = analyze.doc_analyze(b"fake-pdf",effort=effort, # type: ignore[arg-type]- parse_mode=parse_mode, # type: ignore[arg-type]+ parse_mode="auto" if auto_mode else parse_mode, # type: ignore[arg-type])+ assert fake_document.classify.call_count == int(auto_mode)+ getattr(vlm_predictor, vlm_method_name).assert_called_once()assert isinstance(model_json, ModelJson)assert type(model_json.pages) is listassert type(model_json.pages[0]) is listdiff --git a/tests/unittest/test_vlm_remote.py b/tests/unittest/test_vlm_remote.pyindex c7511b4c..ec5e6344 100644--- a/tests/unittest/test_vlm_remote.py+++ b/tests/unittest/test_vlm_remote.py@@ -373,3 +373,38 @@ def test_low_efforts_never_create_remote_client(assert result.effort == effortassert openai_server.requests == []remote_factory.assert_not_called()++[email protected]("tier", ["standard", "advanced"])[email protected]("async_mode", [False, True])[email protected]("enabled", [False, True])+def test_remote_parse_inference_progress(+ openai_server: _OpenAIServer,+ hybrid_stub: None,+ tmp_path: Path,+ monkeypatch: pytest.MonkeyPatch,+ capsys: pytest.CaptureFixture[str],+ tier: str,+ async_mode: bool,+ enabled: bool,+) -> None:+ """真实 PDF 经同步/异步解析和 loopback HTTP 推理后,终端仅出现所属阶段的进度条。"""+ from functools import partial+ from mineru_vl_utils import MinerUClient+ from mineru.model.vlm import runtime++ monkeypatch.setattr(runtime, "MinerUClient", partial(MinerUClient, use_tqdm=enabled))+ monkeypatch.setattr(config.model, "vlm", _settings(openai_server))+ source = _pdf_input(tmp_path)+ result = (+ asyncio.run(parse_async(source, tier=tier, ocr_mode="ocr")) if async_mode else parse(source, tier=tier, ocr_mode="ocr")+ )+ assert "Remote VLM text" in result.markdown()+ stderr = capsys.readouterr().err+ description = "VLM Predict" if tier == "standard" else "Two Step Extraction"+ assert (description in stderr) is enabled+ if enabled:+ assert "100%" in stderr+ if tier == "advanced":+ assert "VLM Predict" not in stderr+ assert any(body is not None for _, _, body in openai_server.requests)